Identification

select:citadl

As described in recon:citadl, the algorithm calculates negative log likelihoods for the electron, pion, kaon, and proton hypothesis for each PWC track. By cutting on differences between the different particle hypotheses, we enforce particle identification.

Kaon identificationKaon identification is straightforward. Since most of the particles produced in target interactions are electrons or pions, we cut on the difference between the kaon and pion hypothesis. (This implies an effective cut on kaon vs. electron since for real kaons, the kaon hypothesis should be closer to the hypothesis than the electron hypothesis.) The cut variable is which is defined as $\wobs{\pion} - \wobs{\kaon}$. We usually require positive values for , meaning that the kaon hypothesis is favored over the pion hypothesis.

Proton identificationProton identification is a little more complex. For we place a strong cut on the variable , defined similarly to . We also place a weaker cut on to remove background from the decays and where the proton is misidentified as a kaon. Typical cut values are $\dwpip >5$ and $\dwkp >1$.

Pion identificationTo identify pions, we use a cut called (pion consistency). This cut determines how likely the pion hypothesis is compared to the other hypotheses. The definition of is

$$\picon \equiv \min (\wobs{\electron},\wobs{\kaon},\wobs{\proton}) - \wobs{\pion}$$ (12)

which allows for values in the range $(-\infty,\infty)$. This type of cut typically incurs small losses in signal at a value of -6, which means that some other hypothesis is favored over the pion hypothesis, but not by a large amount. Because of the limited momentum range of positive pion identification, requiring $\picon > 0$ is inefficient, although the purity of signals with such a cut is very high.